HERES, a Lncrna That Regulates Canonical and Noncanonical Wnt Signaling Pathways Via Interaction with EZH2

Correction

CELL BIOLOGY Correction for “HERES, a lncRNA that regulates canonical and The authors note, “Due to an error during figure preparations, noncanonical Wnt signaling pathways via interaction with EZH2,” the Western blot images for β-actin in Figs. 3E,4H,5C, and 6H by Bo-Hyun You, Jung-Ho Yoon, Hoin Kang, Eun Kyung Lee, appeared incorrectly. We are very sorry for any inconvenience Sang Kil Lee, and Jin-Wu Nam, which was first published No- this Correction may cause.” The full corrected Figs. 3, 4, 5, and 6 vember 15, 2019; 10.1073/pnas.1912126116 (Proc. Natl. Acad. Sci. appear below with their legends. These errors do not affect the U.S.A. 116, 24620–24629). conclusions of the article.

KYSE-30 siControl siHERES_1 siHERES_2 __ __ A B E siControl + 4 CT _ _ _ NC siControl siHERES_1 + + si_1 _ _ _ si_2 siHERES_1 siHERES_2 ++ 3 si_1+pcDNA 0hr _ __ 150 siHERES_2 pcDNA-HERES ++ si_2+pcDNA ** 2 E-cadherin * 100

** N-cadherin CORRECTION

O.D (450nm) 1 ** 50 ** Vimentin * 48hr ** β-actin 0 Wound Closure (%) 0 KYSE-30 KYSE-30 KYSE-30 0hr 24hr 48hr 72hr HCE-7 siControl siHERES_1 siHERES_2 __ __ siControl + 4 CT _ _ _ NC siControl siHERES_1 + + si_1 siHERES_2 _ _ _ 3 si_2 siHERES_1 ++ si_1+pcDNA 0hr _ __ si_2+pcDNA 150 siHERES_2 pcDNA-HERES ++ 2 * 20020.000.0 μmμmμm E-cadherin 100 * N-cadherin

O.D (450nm) 1 * 50 Vimentin 48hr ** ** 0 β-actin Wound Closure (%) 0 0hr HCE-7 HCE-7 HCE-7 24hr 48hr 72hr C D siControl siControl siControl siHERES_1 siHERES_2 siControl siHERES_1 siHERES_2 600 siHERES_1 400 siHERES_1 siHERES_2 300 siHERES_2 400 ** ** 200 ** ** 200 50.050500.000.00 μmμm 300 μm 100

Invading cells 0 0

KYSE-30 KYSE-30 KYSE-30 Number of colonies KYSE-30 siControl siHERES_1 siHERES_2 siControl siControl siHERES_1 siHERES_2 siControl 600 400 siHERES_1 siHERES_1 300 400 siHERES_2 siHERES_2 ** ** 200 200 ** 100 **

200 μm Invading cells 0 0

HCE-7 HCE-7 HCE-7 Number of colonies HCE-7

Fig. 3. HERES modulates cell proliferation, migration, invasion, and colony formation. (A) Cell viability was measured using an MTS assay in KYSE-30 and HCE-7 cells transfected with siControl (NC), siHERES (si_1 and si_2), or siHERES followed by pcDNA-HERES (si_1+pcDNA, si_2+pcDNA). Growth curves were compared between siHERES- and siControl-transfected cells, and between pcDNA-HERES+siHERES- and siHERES-transfected cells. Wound healing assays (B), invasion assays (C), and colony formation assays (D) were performed in KYSE-30 and HCE-7 cells after HERES knockdown (all assay images at 4× magnification). The bar graphs represent the frequency of wound closure (B) and the number of invading cells (C) and colonies formed (D). Data represent the mean ± SD from 3 independent experiments (A–D). (E) Expression of EMT markers in KYSE-30 and HCE-7 cells transfected with siControl or the indicated combinations of siHERES and pcDNA-HERES as determined by immunoblot. *P ≤ 0.05, **P ≤ 0.01.

www.pnas.org PNAS | December 8, 2020 | vol. 117 | no. 49 | 31549–31552 Downloaded by guest on September 26, 2021 CACNA2D3 SFRP2 ACB LDOC1 CACNA1E ) 2 2 DEG CXXC4 ) 20 2 3 SFRP4 N.S. 1 CACNA2D3 2 15 1 0 0 10 -1 BST2 -1 DNA Methylation Expression 5 SLC15A3 (normalized to U6) -2 GRIN1 (HERES-high/-low, log -2 EPSTI1 Expression of HERES BMP7 0 (siHERES/siControl, log -3 Genes -4 -2 0 2 4 RNA Expression Cytosol Nucleus (HERES-high/-low, log2)

D E __ F __ siControl + siControl + chr3: 54,155,000| 54,157,000| 54,159,000| _ _ siHERES_1 _ + _ siHERES_1 + _ _ _ _ CACNA2D3 siHERES_2 + siHERES_2 + 20 CACNA2D3 M H3K4me3 CpG islands 189 CACNA2D3 UM H3K9me3 0.7_ HERES H3K27me3 Methylation 0_ -high LDOC1 M (beta value) 0.7_ H3K36me3 HERES LDOC1 UM 0_ -low H3 300_ HERES H ______0_ -high siControl + + Expression ______300_ siHERES_1 + + + + (read count) HERES ______siHERES_2 ++ ++ 0_ -low ______pcDNA-HERES ++ ++ G CACNA2D3 2000 4000 6000 SFRP2 CACNA2D3 CXXC4 CpG CpG NLK Primer: 1 2 3 4 5 6 7 8 9 10 11 β-catenin 60 siControl C-myc siHERES_1 40 siHERES_2 Cyclin D1

25 MMP7 Fold change **** * Snail 0 ** * * Anti-H3K27me3/IgG 11198765432 10 β-actin Primer KYSE-30 HCE-7

Fig. 4. HERES epigenetically regulates genes involved in canonical and noncanonical Wnt signaling pathways. (A) Changes in the expression of cancer- related genes in response to siHERES treatment compared to siControl are shown. The colored circles indicate genes that are up-regulated (red) or down- regulated (blue) under HERES-depleted conditions. Changes in the expression of the highlighted genes were experimentally confirmed by qRT-PCR. (B) HERES expression in the nuclear and cytoplasmic fractions of KYSE-30 cells as determined by qRT-PCR. (C) Log-scaled fold-changes of expression (x axis) and DNA methylation (y axis) of each gene in the HERES-high versus HERES-low sample groups from the TCGA dataset. The red dots indicate DE genes in ESCC. The highlighted genes are those for which there is anti-correlation between expression and DNA methylation. DEG: DE gene. N.S.: not significant. (D)The CACNA2D3 genomic locus with CpG island tracks, DNA methylation (beta value), and RNA expression (read count) in the HERES-high and HERES-low groups. (E) CACNA2D3 and LDOC1 DNA methylation patterns (methylation [M] and unmethylation [UM]) were measured by MS-PCR in KYSE-30 cells transfected with siControl or siHERES. (F) Immunoblots of histone modification markers in siControl- or siHERES-transfected KYSE-30 cells. (G) ChIP-qPCR analysis of the H3K27me3 level of CACNA2D3 in siControl- or siHERES-transfected KYSE-30 cells. Data represent the mean ± SD from 3 independent experiments. (H)Im- + munoblots of the products of the 3 HERES target genes (CACNA2D3, SFRP2, and CXXC4), components (β-catenin and NLK) of the Wnt/β-catenin and Wnt/Ca2 signaling pathways, and downstream proteins associated with cell proliferation (CMYC and CCND1), invasion (MMP7), and EMT (SNAIL).

31550 | www.pnas.org Downloaded by guest on September 26, 2021 A

B CORRECTION

PNAS | December 8, 2020 | vol. 117 | no. 49 | 31551 Downloaded by guest on September 26, 2021 AB CD

EF G

Fig. 6. HERES regulates CACNA2D3 via interaction with EZH2. (A) Immunoblots of EZH2 and DNMT1 in KYSE-30 cells transfected with either siControl or siHERES. (B) RIP assays were performed with anti-EZH2 in KYSE-30 cell lysates. The quantity of HERES in the cell lysates (input) and the immunoprecipitates was measured by qRT-PCR. (C) IP assays were performed with anti-EZH2 in KYSE-30 cell lysates. The quantity of CACNA2D3 in the cell lysates (input) and the immunoprecipitates was measured by immunoblot. (D) Representation of the WT and mutated (HERES-Mut) HERES sequences used for IP with anti-EZH2. HERES-Mut contains a deletion of the G-rich sequence (index 1) presented in SI Appendix, Fig. S11B.(E) RIP assays were performed with anti-EZH2 in lysates of KYSE-30 cells transfected with either pcDNA-HERES or HERES-Mut (Upper). The bar graph shows the relative amount of HERES after anti-EZH2 IP using lysates of cells transfected with either pcDNA-HERES or HERES-Mut (Lower). (F) RNA FISH (40× magnification) to visualize HERES (red) and FITC staining of CACNA2D3 (green) in KYSE-30 cells transfected with pcDNA (Upper), pcDNA-HERES (Middle), or HERES-Mut (Lower). (Scale bar, 20 μm.) Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) (blue). CACNA2D3 RNA (G) and protein (H) levels were measured in KYSE-30 cells transfected with pcDNA, pcDNA- HERES, or HERES-Mut by qRT-PCR and immunoblot, respectively. Data represent the mean ± SD from 3 independent experiments (B, E, and G). **P ≤ 0.01.

Published under the PNAS license. First published November 16, 2020.

www.pnas.org/cgi/doi/10.1073/pnas.2022237117

31552 | www.pnas.org Downloaded by guest on September 26, 2021 HERES, a lncRNA that regulates canonical and noncanonical Wnt signaling pathways via interaction with EZH2

Bo-Hyun Youa,1, Jung-Ho Yoonb,1, Hoin Kangc, Eun Kyung Leec, Sang Kil Leeb,d,2, and Jin-Wu Nama,e,2 aDepartment of Life Science, College of Natural Sciences, Hanyang University, 04763 Seoul, Republic of Korea; bDivision of Gastroenterology, Department of Internal Medicine, Yonsei Institute of Gastroenterology, Yonsei University College of Medicine, 03722 Seoul, Republic of Korea; cDepartment of Biochemistry, College of Medicine, The Catholic University of Korea, 06591 Seoul, Republic of Korea; dBrain Korea 21 PLUS Project for Medical Science, Yonsei University, 03722 Seoul, Republic of Korea; and eResearch Institute for Convergence of Basic Sciences, Hanyang University, 04763 Seoul, Republic of Korea

Edited by Dennis A. Carson, University of California San Diego, La Jolla, CA, and approved October 17, 2019 (received for review July 18, 2019) Wnt signaling through both canonical and noncanonical pathways overexpression, and dysregulation of regulators, targeting only plays a core role in development. Dysregulation of these pathways the canonical Wnt signaling pathway might not be a universal ther- often causes cancer development and progression. Although the apeutic approach for cancers. Thus, the simultaneous inhibition pathways independently contribute to the core processes, a regula- of aberrant canonical and noncanonical Wnt signaling pathways tory molecule that commonly activates both of them has not yet could also benefit cancer therapy. been reported. Here, we describe a long noncoding RNA (lncRNA), Esophageal squamous cell carcinoma (ESCC), a major histo- HERES, that epigenetically regulates both canonical and noncanonical logical type of primary esophageal cancer in east Asia and other Wnt signaling pathways in esophageal squamous cell carcinoma (ESCC). For this study, we performed RNA-seq analysis on Korean developing countries, is associated with a very poor survival rate – ESCC patients and validated these results on a larger ESCC cohort to that is only 5 15% at 5 y (10, 11), mainly due to delayed diagnosis, identify lncRNAs commonly dysregulated in ESCCs. Six of the dys- a high rate of metastasis, and a lack of effective treatment regulated lncRNAs were significantly associated with the clinical strategies (10–12). Moreover, the benefits of curative surgery for outcomes of ESCC patients and defined 4 ESCC subclasses with advanced stages of ESCC are still unclear (11, 13), and although CELL BIOLOGY different prognoses. HERES reduction repressed cell proliferation, cisplatin-based chemotherapy is commonly used, the effects are migration, invasion, and colony formation in ESCC cell lines and inconsistent among individuals (12, 14). Despite ongoing trials tumor growth in xenograft models. HERES appears to be a transacting with combination therapy, efforts to identify appropriate targets factor that regulates CACNA2D3, SFRP2,andCXXC4 simultaneously to activate Wnt signaling pathways through an interaction with EZH2 via its G-quadruple structure-like motif. Our results suggest Significance that HERES holds substantial potential as a therapeutic target for ESCC and probably other cancers caused by defects in Wnt signaling Aberrant lncRNA expression is responsible for cancer progression pathways. and metastasis, positioning lncRNAs not only as biomarkers but also as promising therapeutic targets for curing cancer. A number epigenetic regulation | long noncoding RNA | Wnt signaling pathway | of lncRNAs have been reported in ESCC but their mechanistic esophageal squamous cell carcinoma roles largely remain unknown. Wnt signaling pathways are often dysregulated in ESCC; however, the role of lncRNAs in such dys- he Wnt signaling pathway is a well-known, evolutionarily regulation was also undetermined. We found 6 lncRNAs that Tconserved pathway that plays important roles in embryonic are significantly dysregulated and correlated with outcomes in development; it has also been widely implicated in numerous ESCC patients. The most upregulated lncRNA, HERES, promotes tumor malignancies (1–4). Wnt signaling can activate both β-catenin– cancer progression and epigenetically regulates canonical and dependent (canonical) and -independent (noncanonical) signal noncanonical Wnt signaling pathways simultaneously through transduction cascades (3, 4). Canonical Wnt signaling results in interaction with EZH2. These results show that HERES repre- translocation of the transcriptional activator β-catenin into the sents an early diagnostic and therapeutic target for squamous- nucleus during embryonic development and cell differentiation cell-type cancers caused by defects in Wnt signaling pathways. (5). Constitutive activation of this pathway by various causes Author contributions: S.K.L. and J.-W.N. designed research; B.-H.Y., J.-H.Y., H.K., and leads to developmental diseases and carcinogenesis (6). In contrast, E.K.L. performed research; B.-H.Y. analyzed data; and B.-H.Y., J.-H.Y., S.K.L., and J.-W.N. noncanonical Wnt pathways are known to be transduced by Wnt wrote the paper. 2+ polarity, Wnt-Ca , and Wnt-atypical protein kinase signaling, The authors declare no competing interest. independent of β-catenin transcriptional activity (7). These path- This article is a PNAS Direct Submission. ways have also been reported to be independently involved in This open access article is distributed under Creative Commons Attribution-NonCommercial- cancer development as well as embryonic development. In par- NoDerivatives License 4.0 (CC BY-NC-ND). ticular, abnormal intracellular levels of the second messenger Data deposition: Raw RNA-seq data and expression tables from ESCC patients have been + Ca2 promote the Wnt signaling pathway, which in turn promotes deposited in the National Center for Biotechnology Information (NCBI) Gene Expression the development and progression of many types of cancers (8). Omnibus (GEO) database, https://www.ncbi.nlm.nih.gov/geo/ (accession no. GSE130078). LncRNA annotation constructed in this study can be downloaded from our website (http:// Controlling Wnt signaling may be a useful strategy for curing big.hanyang.ac.kr/CASOL/catalog.html). cancers caused by aberrations in such signaling. The inhibition of 1B.-H.Y. and J.-H.Y. contributed equally to this work. either aberrant canonical or noncanonical Wnt signaling, how- 2To whom correspondence may be addressed. Email: [email protected] or jwnam@hanyang. ever, has been shown to decrease progression in only a subset of ac.kr. cancers in a context-dependent manner (9). Because aberrations This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. in Wnt signaling pathways result from various causes, such 1073/pnas.1912126116/-/DCSupplemental. as mutations in different Wnt signaling-related genes, ligand www.pnas.org/cgi/doi/10.1073/pnas.1912126116 PNAS Latest Articles | 1of10 to improve the therapy for ESCC have been largely unsuccessful development, the function of a few lncRNAs, including LUCAT1 (15, 16). and CASC9, have been investigated via a candidate-gene ap- Long noncoding RNAs (lncRNAs), defined as transcripts proach (24, 25). Recently, a Chinese group performed high- longer than 200 nt that do not code for functional proteins (17, throughput RNA sequencing (RNA-seq) on tissue from 15 18), have been proposed as regulators of critical biological pro- paired ESCC patients and normal individuals and identified cesses and cancer-related mechanisms (19–21). Because lncRNAs lncRNAs dysregulated in ESCCs (26). Furthermore, they de- can modulate multiple targets at the transcriptional and post- scribed a lncRNA that affects cell proliferation and invasion in transcriptional levels, lncRNAs tend to play functional roles in ESCC cell lines but did not determine a mechanism of action. more than 1 biological pathway. Moreover, mounting evidence Thus, the identification of ESCC-driving lncRNAs and an in- indicates that aberrant lncRNA expression, by modulating cancer- vestigation of their cancer-driving mechanisms have not been related pathways, can be responsible for cancer progression (22, simultaneously carried out. 23). To date, hundreds of lncRNAs have been reported to be Through integrative analyses of ESCC-driving lncRNAs, we dysregulated in cancers and dozens of them have been known to found 6 lncRNAs associated with cancer progression and re- be associated with cancer progression. With respect to ESCC lapse. We also determined that a lncRNA, HERES (highly

YSH Genomic location Localization AB(n=13, paired) C Antisense 1 6 1 Convergent Cytosol 163 14 15 36 Divergent Nucleus 42 43 86 Intergenic 15 Both 113 53 Sense intronic Diff. Loc. 7 1 35 294124 628 Sense overlapping Unidentified Chinese GTEx (n=328) Tandem (n=15, paired) +TCGA (n=95)

D Enhancer overlap E Methyl association F Prognosis 2 4 Overlap 22 High (> 30) Good (P ≤ 0.05) 47 Nonoverlap Middle (≤ 30) Poor (P ≤ 0.05) 55 Low (≤ 10) Not sig. (P > 0.05) 66 36 107

G Class L1 Class L2 Class L3 Class L4 Cohort TCGA YSH Alcohol Yes No NA Tobacco Smoker Reformed Non-smoker NA Reflux Yes No NA Grade G1 G2 G3 GX Tumor stage Stage I Stage II Stage III&IV NA Days OS 2000 100 Vital Alive Dead HERES Expression High Low RP11−1L12.3 CTD−2319I12.1 RP11−114H23.1 RP11−114H23.2 LINC00330

Fig. 1. Characterization of a common set of DE lncRNAs in ESCC. (A) Venn diagram of the DE lncRNAs detected in 3 ESCC cohorts (YSH, Chinese, and GTEx+TCGA). The pie charts (B–F) show the number of DE lncRNAs categorized according to their genomic location (B), subcellular localization (C), enhancer overlap (D), association with DNA methylation (E), and prognostic power (F). (G) The 4 ESCC subclasses based on the 6 prognostic marker lncRNAs in F. The top section presents cohort information, clinical history, pathological features, and survival information from the YSH and TCGA ESCC patients. The various categories are represented as different colors, as shown in the legend on the right. OS is overall survival. The expression patterns of the 6 prognosis-related lncRNAs in the RNA-seq datasets are shown with a colored heatmap in the bottom section (red indicates the top 33% highly expressed lncRNAs associated with a HHR; blue indicates the top 33% highly expressed lncRNAs associated with a LHR).

2of10 | www.pnas.org/cgi/doi/10.1073/pnas.1912126116 You et al. expressed lncRNA in esophageal squamous cell carcinoma), Six lncRNAs Define 4 ESCC Subclasses with Different Clinical Outcomes. promoted tumor progression by regulating components of the To find clinically relevant lncRNAs that are associated with survival canonical and noncanonical Wnt signaling pathways via in- outcomes of patients, Kaplan–Meier survival analysis for all 113 teraction with EZH2, a subunit of polycomb repressive complex DE lncRNAs were performed with TCGA ESCC datasets 2(PRC2). comprising 95 patient samples (Fig. 1F and Dataset S4E). Six DE lncRNAs were significantly associated with survival rates, 2 Results (HERES and RP11-1L12.3) of which were associated with a high ≤ High-Confidence Differentially Expressed (DE) lncRNAs in ESCC. To hazard ratio (HHR; P 0.05) and 4 (RP11-114H23.1, RP11- 114H23.2, CTD-2319I12.1, and LINC00330) of which were asso- identify a set of lncRNAs dysregulated in ESCC, RNA-seq ≤ performed from paired cancerous and noncancerous tissues of ciated with a low hazard ratio (LHR; P 0.05). To delineate how the expression of the 6 lncRNAs stratifies ESCC patients, we 13 ESCC patients in the Yonsei Severance Hospital (YSH) co- clustered samples from the TCGA and YSH cohorts including hort (Dataset S1) and was subjected to transcriptome assembly, additional 10 RNA-seq samples where the clinical values were and lncRNA annotations using our computational pipeline (27) available based on the binary expression patterns (high and low) (SI Appendix,Fig.S1). In total, 6,411 lncRNAs from 4,842 of the 6 lncRNAs, revealing 4 distinct classes of patients: class annotated loci and 1,924 from 1,657 unannotated loci were iden- L1∼L4 (Fig. 1G and Dataset S5). Noticeably, class L1, in which tified (Dataset S2). Of the total of 8,335 lncRNAs, 465 (305 up- only the HHR markers are highly expressed, showed a worse regulated and 160 down-regulated) were significantly dysregulated survival rate than class L3 (P < 0.05) and the other classes (P = in ESCC, exhibiting greater than 2-fold differences in expression, 0.01; Fisher’s exact test). Class L3 tended to display a greater with a false discovery rate (FDR) ≤ 0.01 (Fig. 1A and Dataset S3). overall survival rate than other classes (P < 0.05; Fisher’s exact Then, to identify an ethnically independent set of DE lncRNAs in test). Importantly, class L1 appeared to be significantly asso- ESCC, 113 DE lncRNAs commonly dysregulated in all 3 ESCC ciated with smoking (P < 0.05; Fisher’s exact test), compared to cohorts (YSH, Genotype-Tissue Expression [GTEx] + The Can- other classes. Taken together, these results indicate that the 6 cer Genome Atlas [TCGA], and Chinese) were selected (Fig. 1A). lncRNAs represent prognostic signature genes that can stratify Of the 113 confidence DE lncRNAs, 20 were newly annotated, 32 ESCC patients based on clinical outcomes. were up-regulated, and 81 were down-regulated in ESCC (SI HERES Encodes Alternative Splicing Isoforms, Up-Regulated in ESCCs. Appendix A–C ,Fig.S2 ). A majority of the confidence DE lncRNA Since HERES, 1 of the HHR markers, was greatly up-regulated genes were either located in intergenic regions or were antisense in ESCCs compared to paired adjacent noncancerous samples CELL BIOLOGY to other genes (Fig. 1B); their genomic and clinical features, such (SI Appendix, Fig. S3) and most strongly associated with poor as subcellular localization of the lncRNA (Fig. 1C), associations vital status (Fig. 1G), we investigated whether HERES might be with enhancers (Fig. 1D), and DNA methylation (Fig. 1E), were an ESCC-driving lncRNA. The lncRNA gene encodes 2 isoforms systematically examined (Dataset S4 A–D). As previously reported, with the same transcription start sites (TSSs) (28) and cleavage many overlapped with enhancers (Fig. 1D) and seemed to be and polyadenylation sites (CPSs) (29) (Fig. 2A). Isoform #1, associated with epigenetic markers of other genes (Fig. 1E). HERES.1, is 2,160 nt and contains 2 exons, whereas isoform #2,

ACB chr2: 191,620,000| 191,623,000| 191,626,000| 25 25 CAGE-seq Ex-p1 Het-1A 20 20 KYSE-30 3P-seq Int-p1 15 Int-p2 15 TE-2 FPKM CPC CPAT TE-3 HERES.1 3.82 -0.84 0.005 10 Int-p3 10 HCE-4 HERES.2 1.21 -0.38 0.226 Ex-p2 5 5 HCE-7

Primer: Ex-p2 Int-p3 Int-p2 Int-p1 Ex-p1 (normalized to U6) 0 (normalized to U6) 0 Relative RNA expression Relative RNA expression DEF )

YSH Chinese GTEx+TCGA -2 YSH TCGA 40 6 1.5 100 100 *** *** ** 30 *** 75 75 4 1 Lower (n=31) Lower (n=47) 20 *** 50 50

(FPKM) 2 0.5 Upper (n=35)

10 (stage-free) 25 25 P = 0.045 P = 0.027 Upper (n=48)

0 Percent survival (%) 0 0 Expression of HERES 0

Expression of HERES 0

(normalized to U6, x10 0 0 500 500 1000 1500 2000 1000 1500 2000 Cancer=23) Cancer=15) Normal=328) Cancer=95) Cancer=66) Normal=21) (n (n n (n =66) (n (n =66) Days Days ( n n Non-cancer=23) Non-cancer=15) Non-cancer( Non-cancer( (n (n

Fig. 2. HERES is a highly expressed lncRNA in ESCC. (A) The HERES genomic locus with CAGE-seq and 3P-seq signals. qRT-PCR primer sets were designed to recognize exonic (2) and intronic regions (3). The coding potentials calculated by CPC and CPAT are indicated on the right. (B) qRT-PCR results using the 5 primer sets in KYSE-30 cells. (C) The HERES expression level was measured in 5 ESCC cell lines and a normal esophageal cell line (Het-1A). (B and C) Error bars represent the mean ± SD from 3 independent experiments. (D) The box plots show the HERES expression levels in normal, noncancerous, and cancerous tissues from the YSH cohort (paired), the Chinese cohort (paired), and the TCGA cohort. (E) HERES expression levels measured by qRT-PCR in additional frozen tissue samples including YSH ESCC (n = 66) and adjacent samples (noncancer; n = 66) (Left) and in normal mucosa tissues (n = 21) from reflux symptom patients (Right). (F) Survival analyses of YSH and TCGA patients from whom the ESCC samples were obtained based on the HERES expression level. **P ≤ 0.01, ***P ≤ 0.001.

You et al. PNAS Latest Articles | 3of10 HERES.2, is an intron-retained, single-exonic transcript that is expression changes were further confirmed in 66 ESCC samples 6675 nt. Both isoforms were confirmed to lack coding potential from the YSH cohort using qRT-PCR, which revealed an elevated (Fig. 2A). Analysis of ESCC RNA-seq datasets (SI Appendix, Fig. level of HERES in cancers compared to adjacent noncancerous S4A) and qRT-PCR in an ESCC cell line (KYSE-30) (Fig. 2 A samples (Fig. 2 E, Left). HERES expression in the adjacent non- and B) showed that HERES.1 is the major isoform. Only a short cancerous samples was higher than that in normal mucosa from the region in HERES exon 1 displays sequence conservation with a normal population (Fig. 2 E, Right). As observed in the YSH cohort region in the mouse genome; the intergenic HERES locus be- (Fig. 2 F, Left and Dataset S6), HERES levels were significantly tween the GLS and NAB1 genes on chromosome 2 shows syn- correlated with stage-free survival rates in the TCGA ESCC cohort teny conservation in mouse (SI Appendix, Fig. S4B). (Fig. 2 F, Right), and multivariate analysis with clinical information Elevated HERES expression was then validated in other revealed that the HERES level was strongly associated with tumor ESCC cell lines. Compared to that in a normal esophageal epithelial grade (P = 0.004; Fisher’s exact test) but not with other clinico- cell line (Het-1A), HERES expression appeared to be up-regulated pathological factors (SI Appendix,Fig.S4D). greater than 10-fold in all tested ESCC cell lines (Fig. 2C), as observed in ESCC samples (Fig. 2D). HERES was significantly up- HERES Promotes Cell Proliferation, Migration, Invasion, and Colony regulated not only in ESCC, but also in esophageal adenocarci- Formation. To investigate whether HERES is involved in cancer noma (ESAD) and other squamous carcinomas (head and neck, development and progression, the effects of HERES knockdown and lung squamous cell carcinoma; HNSC and LUSC), but not on cell proliferation, migration, invasion, and colony formation were in lung adenocarcinoma (LUAD) (SI Appendix, Fig. S4C). These explored with siControl- and siHERES-treated cells. Introduction

A B E KYSE-30 siControl siHERES_1 siHERES_2 _ _ __ siControl + 4 CT _ _ _ NC siControl siHERES_1 + + si_1 _ _ _ si_2 siHERES_1 siHERES_2 + + 3 si_1+pcDNA 0hr _ _ _ 150 siHERES_2 pcDNA-HERES ++ si_2+pcDNA ** 2 E-cadherin * 100 ** N-cadherin

O.D (450nm) 1 ** 50 ** Vimentin * 48hr ** β-actin 0 Wound Closure (%) 0 KYSE-30 KYSE-30 KYSE-30 0hr 24hr 48hr 72hr HCE-7 siControl siHERES_1 siHERES_2 _ _ __ siControl + 4 CT _ _ _ NC siControl siHERES_1 + + si_1 siHERES_2 _ _ _ 3 si_2 siHERES_1 + + si_1+pcDNA 0hr _ _ _ si_2+pcDNA 150 siHERES_2 pcDNA-HERES ++ 2 * 20020.00.00 μmμμmm E-cadherin 100 * N-cadherin

Invading cells 0 0

200 μm Invading cells 0 0

HCE-7 HCE-7 HCE-7 Number of colonies HCE-7

Fig. 3. HERES modulates cell proliferation, migration, invasion, and colony formation. (A) Cell viability was measured using an MTS assay in KYSE-30 and HCE-7 cells transfected with siControl (NC), siHERES (si_1 and si_2), or siHERES followed by pcDNA-HERES (si_1+pcDNA, si_2+pcDNA). Growth curves were compared between siHERES- and siControl-transfected cells, and between pcDNA-HERES+siHERES- and siHERES-transfected cells. Wound healing assays (B), invasion assays (C), and colony formation assays (D) were performed in KYSE-30 and HCE-7 cells after HERES knockdown (all assay images at 4× magnification). The bar graphs represent the frequency of wound closure (B) and the number of invading cells (C) and colonies formed (D). Data represent the mean ± SD from 3 independent experiments (A–D). (E) Expression of EMT markers in KYSE-30 and HCE-7 cells transfected with siControl or the indicated combinations of siHERES and pcDNA-HERES as determined by immunoblot. *P ≤ 0.05, **P ≤ 0.01.

4of10 | www.pnas.org/cgi/doi/10.1073/pnas.1912126116 You et al. of an siRNA targeting HERES (siHERES_1 or siHERES_2) to (Fig. 4E and SI Appendix,Fig.S7F–L). On the other hand, although KYSE-30 and HCE-7 cells significantly reduced HERES expression SFRP2 and CXXC4 did not display DNA methylation changes in the (SI Appendix, Fig. S5). The proliferation indices (optical density analysis of the TCGA ESCC samples, the expression and DNA [O.D.] values) were significantly reduced in both siHERES_1 methylation signals of the Wnt signaling-related genes were changed and siHERES_2-treated cells compared to siControl-treated similarly to those of CACNA2D3 and LDOC1 in the siHERES- cells (Fig. 3A), and the introduction of a HERES pcDNA ex- treated cells compared to the siControl-treated cells (SI Appendix, pression construct partly rescued the proliferation activity Fig. S7B and S8 A–C). (Fig. 3A), indicating that HERES can regulate cell proliferation. Because DNA methylation is often associated with histone Migration and invasion assays showed that both cell migration modifications (35, 36), global changes in histone modification and invasion were greatly reduced in siHERES-treated cells markers in response to HERES knockdown were examined, re- compared to siControl-treated cells (Fig. 3 B and C). In addition, vealing a marked decrease in H3K27me3 levels (Fig. 4F). We then HERES knockdown also reduced colony formation measured at investigated where the H3K27me3 signal was depleted in the 14 d after siRNA transfection (Fig. 3D). A role for HERES in genomic regions of 3 Wnt signaling pathway genes in siHERES- malignant ESCC progression was confirmed by the reduction of treated cells using chromatin immunoprecipitation (ChIP)-qPCR N-cadherin and vimentin levels in siHERES-treated cells and by analysis. Significantly reduced H3K27me3 signals were observed the rescue of these levels by the introduction of the HERES at specific sites in the genes (recognized by primer 5 for CACNA2D3, pcDNA construct to the cells (Fig. 3E and SI Appendix, Fig. S6). primers 3, 4, and 5 for SFRP2, and primers 9 and 10 for These results suggest that HERES can promote cancer progression CXXC4) in siHERES-depleted cells (Fig. 4G and SI Appendix, and metastasis. Fig. S8 D and E). Previous studies reported that CACNA2D3 down-regulation + HERES Affects Wnt Signaling Pathway-Related Genes. To study the inhibited the noncanonical Wnt/Ca2 signaling pathway by de- means by which HERES promotes cancer development and pro- creasing the intracellular calcium level and NLK expression (32) gression, the changes in expression of ∼730 cancer-related pathway and that SFRP2 and CXXC4 play roles as negative regulators of genes were analyzed in siHERES-treated and siControl-treated the canonical Wnt signaling pathway (34, 37). We thus examined KYSE-30 cells using the NanoString nCounter PanCancer Path- changes in the expression of 2 Wnt signaling-related factors, ways Panel (Fig. 4A and Dataset S7). Seventy-seven cancer-related NLK and β-catenin, in siHERES-treated cells (Fig. 4H and SI genes (34 for up-regulation and 43 for down-regulation) were Appendix, Fig. S9). As expected, HERES reduction increased the dysregulated greater than 2-fold in siHERES-treated cells NLK level and decreased β-catenin in KYSE-30 and HCE-7 compared to siControl cells (Fig. 4A); the expression changes of cells. In addition, changes in the expression of Wnt downstream CELL BIOLOGY the 2 most up-regulated genes (CACNA2D3 and SFRP2) and the targets were also confirmed (Fig. 4H and SI Appendix,Fig.S9). In 2 most down-regulated genes (BMP7 and GRIN1) in this group contrast, introducing the pcDNA-HERES construct to cells were confirmed by qRT-PCR (SI Appendix, Fig. S7 A–D). No- reverted the expression levels of NLK, β-catenin, and Wnt ticeably, among the genes dysregulated by HERES reduction, 14 downstream targets (Fig. 4H and SI Appendix,Fig.S9). Taken belong to the Wnt signaling pathway, and half of the 10 most up- together, these results suggest that HERES down-regulation in regulated genes (CACNA2D3, SFRP2, CACNA1E, CXXC4, and cancers perturbs and promotes canonical and noncanonical SFRP4) are involved in the Wnt signaling pathway (Dataset S7). Wnt signaling pathways via epigenetic regulation, resulting in CACNA2D3, which encodes a subunit of the calcium channel the inhibition of cancer progression. protein complex, was previously shown to be induced in ESCC (30) and other cancers (31, 32) via epigenetic mechanisms, and Inhibition of HERES Arrests the Cell Cycle at the G1/S Checkpoint and + its down-regulation led to inactivation of Wnt/Ca2 signaling Induces Apoptosis. Because 2 of the downstream targets of pathway (32). SFRP2 encodes a member of the SFRP family that HERES, CCND1 and CACNA2D3, are known to regulate the modulates the Wnt signaling pathway; SFRP2 hypermethylation cell cycle and apoptosis (30, 32), the effect of the loss of HERES is known to enhance cell invasiveness in both cancers and non- on the cell cycle and apoptotic processes was examined. Cell cancerous diseases (33, 34). The enrichment of canonical and counting showed that siHERES-treated cells were arrested at noncanonical Wnt signaling pathway-related genes among the G0/G1 (Fig. 5A). Flow cytometry showed that siHERES-treated genes that respond to HERES depletion, together with results cell populations exhibited significantly increased levels of apo- from previous studies, suggest that HERES may regulate cancer ptosis compared to siControl-treated cells (Fig. 5B). An induction development via control of Wnt signaling pathways. of apoptotic factors, such as cleavage of poly (ADP ribose) poly- merase (PARP), cleaved caspase-9, and Bax, and a reduction of the HERES Regulates Wnt Signaling Pathways at the Epigenetic Level. As anti-apoptotic factor, Bcl-2, were confirmed in siHERES-treated HERES appeared to be enriched in the nucleus rather than the cells. However, the rescue of HERES expression reverted the ex- cytoplasm (Fig. 4B and SI Appendix,Fig.S2C)andnucleus- pression of these factors to levels in control cells (Fig. 5C). localized lncRNAs are often reported to be epigenetic regulators, a potential epigenetic role for HERES was investigated by an- HERES Interacts with EZH2 to Regulate CACNA2D3. We then asked alyzing publicly available array-based DNA methylation and how HERES regulates the expression of target genes at the RNA-seq data from TCGA ESCC samples. Based on their HERES epigenetic level. To address this question, binding sites for possible expression level, the ESCC samples were first divided into 2 sub- epigenetic modulators that can drive the histone methylation of groups, HERES-high and HERES-low, and the changes in expres- target genes were first examined using publicly available ChIP- sion and DNA methylation of the protein-coding genes were then seq datasets from the ENCODE project. We found that all 3 compared between the subgroups (Fig. 4C and Dataset S8). Of the HERES target genes contained enhancer of EZH2 binding sites genes, CACNA2D3 and LDOC1 (Fig. 4D and SI Appendix,Fig.S7E) in their promoter regions (SI Appendix,Fig.S10). Because EZH2, were down-regulated and hypermethylated in the HERES-high a subunit of the PRC2, has a well-known role in histone methyl- group, whereas EPSTI1, SLC15A3,andBST2 were up-regulated ation to generate H3K27me3 and is known to interact with nuclear and hypomethylated in the HERES-high subgroup. The expression lncRNAs (38), we suspected that EZH2 would be a binding part- and DNA methylation changes were confirmed in HERES-depleted ner of HERES. To examine the molecular relationship between KYSE-30 cells by qRT-PCR and methylation-specific (MS) PCR. HERES and EZH2, EZH2 RNA and protein levels were quanti- Only 2 down-regulated genes (CACNA2D3 and LDOC1) were fied in siControl- and siHERES-treated KYSE-30 cells, showing confirmed to have both expression and DNA methylation changes that HERES reduction decreased the EZH2 protein level but not

You et al. PNAS Latest Articles | 5of10 CACNA2D3 A SFRP2 C B LDOC1 CACNA1E ) 2 2 DEG CXXC4 ) 20 2 3 SFRP4 N.S. 1 CACNA2D3 2 15 1 0 0 10 /siControl, log -1 BST2 -1 DNA Methylation Expression 5 SLC15A3 (normalized to U6) -2 GRIN1 (HERES-high/-low, log -2 EPSTI1 Expression of HERES BMP7 0 (siHERES -3 Genes -4 -2 0 2 4 RNA Expression Cytosol Nucleus (HERES-high/-low, log2) D E F _ _ _ _ siControl + siControl + chr3: 54,155,000| 54,157,000| 54,159,000| _ _ siHERES_1 _ + _ siHERES_1 + _ _ _ _ CACNA2D3 siHERES_2 + siHERES_2 + 20 CACNA2D3 M H3K4me3 CpG islands 189 CACNA2D3 UM H3K9me3 0.7_ HERES H3K27me3 Methylation 0_ -high LDOC1 M (beta value) 0.7_ H3K36me3 HERES LDOC1 UM 0_ -low H3 300_ HERES H ______0_ -high siControl + + Expression ______300_ siHERES_1 + + + + (read count) HERES ______siHERES_2 + + + + 0_ -low ______pcDNA-HERES ++ ++ CACNA2D3 G 2000 4000 6000 SFRP2 CACNA2D3 CXXC4 CpG CpG NLK Primer: 1 2 3 4 5 6 7 8 9 10 11 β-catenin 60 siControl C-myc siHERES_1 40 siHERES_2 Cyclin D1

25 MMP7 Fold change **** * Snail 0 ** * * Anti-H3K27me3/IgG 1112 3 4 5 6 7 8 9 10 β-actin Primer KYSE-30 HCE-7

Fig. 4. HERES epigenetically regulates genes involved in canonical and noncanonical Wnt signaling pathways. (A) Changes in the expression of cancer- related genes in response to siHERES treatment compared to siControl are shown. The colored circles indicate genes that are up-regulated (red) or down- regulated (blue) under HERES-depleted conditions. Changes in the expression of the highlighted genes were experimentally confirmed by qRT-PCR. (B) HERES expression in the nuclear and cytoplasmic fractions of KYSE-30 cells as determined by qRT-PCR. (C) Log-scaled fold-changes of expression (x axis) and DNA methylation (y axis) of each gene in the HERES-high versus HERES-low sample groups from the TCGA dataset. The red dots indicate DE genes in ESCC. The highlighted genes are those for which there is anti-correlation between expression and DNA methylation. DEG: DE gene. N.S.: not significant. (D)The CACNA2D3 genomic locus with CpG island tracks, DNA methylation (beta value), and RNA expression (read count) in the HERES-high and HERES-low groups. (E) CACNA2D3 and LDOC1 DNA methylation patterns (methylation [M] and unmethylation [UM]) were measured by MS-PCR in KYSE-30 cells transfected with siControl or siHERES. (F) Immunoblots of histone modification markers in siControl- or siHERES-transfected KYSE-30 cells. (G) ChIP-qPCR analysis of the H3K27me3 level of CACNA2D3 in siControl- or siHERES-transfected KYSE-30 cells. Data represent the mean ± SD from 3 independent experiments. (H) Immunoblots of the products of the 3 HERES target genes (CACNA2D3, SFRP2, and CXXC4), components (β-catenin and NLK) of the Wnt/β-catenin and Wnt/Ca2+ signaling pathways, and downstream proteins associated with cell proliferation (CMYC and CCND1), invasion (MMP7), and EMT (SNAIL).

6of10 | www.pnas.org/cgi/doi/10.1073/pnas.1912126116 You et al. KYSE-30 A siControl siHERES_1 siHERES_2 800 siControl 500 500 50 siHERES_1 400 600 400 40 siHERES_2 300 300 30 400 * * ** * Count Count Count 200 200 20 * * * 200 Cell count 100 100 10 0 0 0 0 50K 100K 150K 200K 50K 100K 150K 200K 50K 100K 150K 200K Sub-G0 G1 S G2 Propidium Iodide-A Propidium Iodide-A Propidium Iodide-A HCE-7 siControl siHERES_1 siHERES_2 siControl 800 800 1000 50 siHERES_1 800 600 600 40 siHERES_2 600 30 400 400 ** * Count Count Count 400 20

200 200 Cell count 200 10 * * 0 0 0 0 50K 100K 150K 200K 50K 100K 150K 200K 50K 100K 150K 200K Sub-G0 G1 S G2 Propidium Iodide-A Propidium Iodide-A Propidium Iodide-A

KYSE-30 B siControl siHERES_1 siHERES_2 105 105 105 50 ** siControl 104 104 104 40 ** siHERES_1 siHERES_2 30 103 103 103 20 CELL BIOLOGY 102 102 102 10 0 0 0 -102 -102 -102 Apoptotic rate (%) 0 010103 104 5 010103 104 5 010103 104 5 Propidium Iodide-A::Propidium Iodide-A Propidium Iodide-A::Propidium Iodide-A Propidium Iodide-A::Propidium Iodide-A APC-A::APC-A APC-A::APC-A APC-A::APC-A HCE-7 siControl siHERES_1 siHERES_2

105 105 105 80 siControl ** 4 4 4 ** siHERES_1 10 10 10 60 siHERES_2 103 103 103 40

102 102 102 20 0 0 0 -102 -102 -102 Apoptotic rate (%) 0 010103 104 5 010103 104 5 010103 104 5 Propidium Iodide-A::Propidium Iodide-A Propidium Iodide-A::Propidium Iodide-A Propidium Iodide-A::Propidium Iodide-A APC-A::APC-A APC-A::APC-A APC-A::APC-A

C siControl + _ _ __ siControl + _ _ _ siHERES_1 _ + _ + _ siHERES_1 _ + _ + ______siHERES_2 + + MWM siHERES_2 + + MWM ______pcDNA-HERES ++(kDa) pcDNA-HERES ++(kDa) PARP -116 PARP -116 -89 -89 Cleaved Caspase-9 -37 Cleaved Caspase-9 -37 Bcl-2 -26 Bcl-2 -26 Bax -23 Bax -23 Cytochrome C -15 Cytochrome C -15 Cleaved Caspase-3 -15 Cleaved Caspase-3 -15 β-actin -45 β-actin -45 KYSE-30 HCE-7

Fig. 5. The effect of HERES on the cell cycle and apoptosis. Cell cycle (A) and apoptosis (B) assays were performed on siRNA-transfected KYSE-30 and HCE-7 cells. (A) Cell cycle analysis of siRNA-transfected KYSE-30 and HCE-7 cells by flow cytometry. The bar graph shows the percentage of cells in sub-G0, G1, S, and G2 phases in siRNA-transfected KYSE-30 and HCE-7 cell populations. (B) Apoptosis was measured by flow cytometry using PI/Annexin V staining. The bar graph represents the percentage of apoptotic cells in each population. Data represent the mean ± SD from 3 independent experiments. (C) Apoptosis markers were assessed by immunoblot in KYSE-30 and HCE-7 cells transfected with siControl or siHERES and/or pcDNA-HERES. *P ≤ 0.05, **P ≤ 0.01. the RNA level (Fig. 6A and SI Appendix,Fig.S11A). Subsequently, EZH2. To validate a direct interaction between HERES and RNA immunoprecipitation (RIP) (Fig. 6B) and EZH2 IP (Fig. 6C) PRC2-EZH2, we then searched for PRC2-EZH2 binding motifs assays showed the interaction of HERES and CACNA2D3 with in the HERES sequence. Because the PRC2 complex including

You et al. PNAS Latest Articles | 7of10 ADB C KYSE-30 _ _ _ 4 IP:EZH2 + siControl + HERES IB:EZH2 siHERES_1 _ + _ 3 GAPDH HERES.1 _ _ siHERES_2 + IB:IgG HERES.2 2 _ EZH2 IP:EZH2 + DNMT1 1 CACNA2D3 WT: TATGGAAAAGGTGGTGGAGGTAGAAAGAC HERES-Mut: TATGGAAAAAGAAAGAC β-actin GAPDH Relative RNA quantity 0 (normalized to GAPDH) _ _ KYSE-30 EZH2: + + KYSE-30 E pcDNA- pcDNA- F G HERES Mut HERES Mut DAPI HERES CACNA2D3 Merge pcDNA pcDNA-HERES HERES 4 HERES-Mut pcDNA 18S rRNA ** 3 IP:IgG IP:EZH2 2 pcDNA-HERES HERES-Mut 1 2.0 pcDNA-HERES ** ** 0

(normalized to GAPDH) KYSE-30 1.5 Expression of CACNA2D3 H pcDNA- 1.0 pcDNA HERES Mut 0.5 HERES-Mut CACNA2D3 0 Relative RNA quantity 20.0 μm β-actin

(normalized to 18S rRNA) IgG EZH2 KYSE-30 KYSE-30 KYSE-30

Fig. 6. HERES regulates CACNA2D3 via interaction with EZH2. (A) Immunoblots of EZH2 and DNMT1 in KYSE-30 cells transfected with either siControl or siHERES. (B) RIP assays were performed with anti-EZH2 in KYSE-30 cell lysates. The quantity of HERES in the cell lysates (input) and the immunoprecipitates was measured by qRT-PCR. (C) IP assays were performed with anti-EZH2 in KYSE-30 cell lysates. The quantity of CACNA2D3 in the cell lysates (input) and the immunoprecipitates was measured by immunoblot. (D) Representation of the WT and mutated (HERES-Mut) HERES sequences used for IP with anti-EZH2. HERES-Mut contains a deletion of the G-rich sequence (index 1) presented in SI Appendix, Fig. S11B.(E) RIP assays were performed with anti-EZH2 in lysates of KYSE-30 cells transfected with either pcDNA-HERES or HERES-Mut (Upper). The bar graph shows the relative amount of HERES after anti-EZH2 IP using lysates of cells transfected with either pcDNA-HERES or HERES-Mut (Lower). (F) RNA FISH (40× magnification) to visualize HERES (red) and FITC staining of CACNA2D3 (green) in KYSE-30 cells transfected with pcDNA (Upper), pcDNA-HERES (Middle), or HERES-Mut (Lower). (Scale bar, 20 μm.) Nuclei were stained with 4′,6- diamidino-2-phenylindole (DAPI) (blue). CACNA2D3 RNA (G) and protein (H) levels were measured in KYSE-30 cells transfected with pcDNA, pcDNA-HERES, or HERES-Mut by qRT-PCR and immunoblot, respectively. Data represent the mean ± SD from 3 independent experiments (B, E, and G). **P ≤ 0.01.

EZH2 is known to be recruited by G-rich motifs (39), we scanned and its target genes, finding that the reduction of HERES was for G-rich regions in HERES transcripts, leading to the identification maintained for 4 wk after siHERES injection (SI Appendix, Fig. of 6 regions including 2 potential g-quadruple structure motifs (SI S12), whereas the levels of HERES targets were significantly Appendix,Fig.S11B). EZH2-IP and qRT-PCR showed that a single increased in tumor samples derived from HERES-depleted cells region with 4 GGW repeats (index 1) was significantly enriched in compared to control cells (Fig. 7D). We also confirmed that EZH2-IP (SI Appendix,Fig.S11C). To further investigate if the global H3K27me3 and EZH2 levels were decreased in HERES- HERES GGW repeat sequence (index 1) is necessary for an in- depleted tumor samples (Fig. 7D), suggesting that HERES is a teraction with EZH2, we constructedaplasmidvectorthatharbors promising candidate therapeutic target that controls tumor a HERES sequence that lacks the GGW repeat region (HERES- growth through the regulation of canonical and noncanonical Mut) (Fig. 6D). A RIP assay confirmed that EZH2 failed to Wnt signaling pathways in vivo (Fig. 7E). interact with HERES-Mut in KYSE-30 cells (Fig. 6E). RNA fluorescence in situ hybridization (FISH) of HERES and fluo- Discussion rescein isothiocyanate (FITC) staining of CACNA2D3 validated A series of computational and experimental analyses showed that HERES was principally localized to the nucleus and that the that HERES transcriptionally controls multiple target genes in GGW sequence (index 1) is necessary for the interaction with the Wnt signaling pathways at the epigenetic level by interacting EZH2 to down-regulate CACNA2D3 (Fig. 6F). Cells transfected with the EZH2-PRC2 complex. Because the targets and HERES with HERES-Mut exhibited significantly increased CACNA2D3 are generally located on different chromosomes, HERES appears at both the RNA and protein level, whereas HERES overexpression (pcDNA-HERES) reduced the CACNA2D3 level to act via EZH2-PRC2 in trans rather than in cis.Intriguingly, (Fig. 6 G and H). HERES RNA contains some repeat elements including GGW and Alu repeats, which extensively match sequences in the up- HERES as a Candidate Therapeutic Target. To investigate whether stream regions of the target genes including CACNA2D3 (SI HERES controls tumor growth in vivo, we carried out xenograft Appendix, Fig. S13). Particularly, the Alu repeats would provide assays with siControl- and siHERES-treated cancer cell lines complementary base-pairing between HERES its target DNA (Fig. 7). Both the volume and weight of tumors derived from sequences as well as it might be related to the nuclear localiza- HERES-depleted samples were significantly reduced compared tion of HERES, as previously reported (40). to tumors derived from control cells 4 wk after the injection (Fig. Although we reported in this study that a G-rich motif in 7 A–C). We further examined changes in the expression of HERES HERES is important for binding to EZH2, it remains unclear

8of10 | www.pnas.org/cgi/doi/10.1073/pnas.1912126116 You et al. A B E PRC2 1000 siControl (n=6) SUZ12 ) 3 siHERES (n=6) * EZH2 EED 800 (SFRP2) SFRP HERES 600 siControl 400 * Wnt (CACNA2D3) 200 CACNα2δ3

Tumor volume (mm Membrane

0 siHERES (CXXC4) Idax Ca2+

0 week 1 week 2 week 3 week 4 week _ _ β-catenin Dvl CaMKII C DsiControl + + _ _ siControl (n=6) siHERES_1 + + siHERES (n=6) Cytoplasm Nucleus 0.4 CACNA2D3 β-catenin NLK 0.3 SFRP2 TCF/LEF factors * CXXC4 0.2 WNT ON H3K27me3 0.1

Tumor weight (g) EZH2 0 Proliferation (CMYC, CCND1) β-actin Metastasis (MMP7) EMT (SNAIL) siControlsiHERES Mice #1 #2

Fig. 7. Expression of HERES regulates tumorigenicity in xenograft models. KYSE-30 cells transfected with either siControl or siHERES were injected into nude mice (6 mice for each group). The resulting xenograft tumor volumes (A and B) and weights (C) are shown. (A) Tumor growth curves showing that tumors in the siHERES group grew markedly slower than those in the siControl group. (B) Images of tumor volumes from the xenograft models. (C) Tumor weights in the siHERES and siControl groups 4 wk after cell injection. Data represent the mean ± SD. (D) Immunoblot analysis (#1 and #2) of levels of key components (CACNA2D3, SFRP2, and CXXC4) from the canonical and noncanonical Wnt signaling pathways and of H3K27me3 and EZH2 in the siHERES and siControl xenograft models. (E) A graphic illustration of HERES-regulated canonical and noncanonical Wnt signaling pathways in ESCC. *P ≤ 0.05. which part of EZH2 interacts with HERES. A previous study RNAlater (Ambion,) and then stored at −80 °C. Total RNA was extracted CELL BIOLOGY showed that the N-terminal region of EZH2 is important for from ESCC cells and tissues and then subjected to qRT-PCR for construction RNA binding through a G-rich motif (41). A series of deletion of RNA-seq libraries (SI Appendix, Methods for more details). All tissue mutants of human PRC2 revealed that the basic N-terminal helix samples were obtained after receiving written informed consent from pa- – tients according to the Declaration of Helsinki, and this study was approved of EZH2, particularly residues 32 42 in the helix, are the most by Institutional Review Board (IRB) of the Yonsei University College of Medicine critical for RNA binding through a G-rich motif. Given these (# 4–2011-0891). results, the G-rich motif embedded in HERES probably also interacts with the basic N-terminal helix of EZH2, although such Annotations and Expression Profiling of lncRNAs. To profile both annotated a direct interaction needs to be verified. and unannotated lncRNAs expressed in ESCC, transcriptome assembly and Transcription factor ChIP-seq data from the ENCODE proj- lncRNA annotations were performed over the initial 13 RNA-seq datasets ect revealed that the first HERES exon contains some enhancer- obtained from the YSH cohort using our computational pipeline (SI Appendix, related transcription factor binding sites (TFBSs) for CEBPB, Methods for more details). Using the resulting annotations of all lncRNAs, EP300, and AP-1 subunits (JUN, FOS) (SI Appendix, Fig. S14), lncRNA expression levels were measured over the YSH cohort, publicly avail- suggesting that HERES could be regulated epigenetically by able ESCC cohorts (95 tumor samples from TCGA and 15 paired samples from modulation of the chromatin state at its locus. On the other a Chinese ESCC cohort) (26, 43), and GTEx Esophagus mucosa datasets (328 samples) (44) (SI Appendix, Methods for more details). Because the RNA-seq hand, the expression of HERES near enhancer-related TFBSs data from the TCGA ESCC and GTEx Esophagus mucosa datasets were raises the possibility that HERES is an enhancer RNA (eRNA) unstranded types (reads with no strand information), the strand information that regulates neighboring genes in cis. However, 2 following was predicted and the unstranded reads were converted to RPDs (reads with observations argue against this idea: first, HERES is highly predicted directions) using CAFE (27). abundant and includes both a 5′ cap and 3′ polyadenylation, unlike eRNAs, and second, the genes neighboring HERES were Methylation-Specific PCR (MS-PCR). Genomic DNA was extracted from KYSE- only marginally affected by siHERES transfection (SI Appendix, 30 cells using a DNeasy Blood and Tissue kit (Qiagen). An EZ DNA Methylation- Fig. S15). Gold kit (ZymoResearch) was used for DNA bisulfite transformation. MS-PCR Although there have been reports that some lncRNAs partici- was performed using primers specific for methylated and unmethylated genes pate in regulating the Wnt signaling pathway, their targets appear (Dataset S9). limited (42). Our results suggest that HERES could be a master ChIP Assay. For chromatin shearing in lysates of cells transfected with small regulator of the Wnt signaling pathway, because it controls key 2+ interfering RNA (siRNA) (siHERES or siControl), we used waterbath sonication components of both canonical and Ca -related noncanonical for 30 cycles under cooling conditions, 15 s on, 30 s off (170–190 W). The pathways (Fig. 7E). Our results highlight the potential significance fragmented chromatin was extracted using a High-Sensitivity ChIP kit of HERES in terms of targeted therapy, where it could be used to (ab185913; Abcam), according to the manufacturer’s protocol. A total of 2+ manipulate Wnt signaling pathways and Ca homeostasis. 5 μg of total chromatin was used for ChIP with anti-H3K27me3 (ab6002; Abcam) and the mock immunoprecipitation (IP) (IgG, ab185913; Abcam) at Materials and Methods 4 °C overnight. After cross-link reversal and DNA purification, 1 μL of eluted Sample Collection and Preparation. Samples from 23 Korean ESCC patients DNA was used for qRT-PCR with target region primers. The primer sequences were used for RNA-seq; an additional 43 ESCC patients were also enrolled in for qRT-PCR are shown in Dataset S9. this study. Clinicopathological findings and tissues were obtained pro- spectively (Dataset S6). Cancerous and adjacent noncancerous mucosa tissues RIP Assay. To immunoprecipitate RNA-protein complexes, cells were first lysed were collected by endoscopic biopsy from all 66 ESCC patients, regarded as with nuclear isolation buffer (1.28 M sucrose, 40 mM Tris·HCl pH 7.5, 20 mM the YSH (Yonsei Severance Hospital) cohort. Additionally, normal mucosal MgCl2, 4% Triton X-100) and resuspended in RIP buffer (150 mM KCl, 25 mM tissues from 21 patients with reflux symptoms, as well as ESCC cell lines, were Tris·HCl pH 7.4, 5 mM EDTA, 0.5 mM DTT, 0.5% Nonidet P-40, 1 U/μL RNase used in the study. Immediately after collection, tissues were transferred to inhibitor, and protease inhibitor). To achieve chromatin shearing, waterbath

You et al. PNAS Latest Articles | 9of10 sonication was administered for 30 cycles and the lysate was cleared by of siControl or siHERES, 2 × 106 KYSE-30 cells were suspended in 200 μLof centrifugation. The appropriate antibody (2 μg) was added to the resulting Matrigel solution (BD Biosciences) and s.c. implanted onto the posterior supernatant (600–800 μg) and the mixture was carefully incubated at 4 °C for flank of BALB/c nude mice (6-wk-old male, n = 6). The volume of the 24 h or more on a rotator. After incubation, 20 μL of Magna Chip protein resulting cell mass after 4 wk was measured using a vernier caliper and the magnetic beads (Merck Millipore) were added, and the solution rotated at mass of each xenograft tumor was analyzed after the animal was euthanized 4 °C for 2 h. After washing with RIP buffer, qRT-PCR and Western blotting according to the Guide for the Care and Use of Laboratory Animals (45). were carried out following RNA purification with TRIzol reagent or SDS gel electrophoresis, respectively. Data Availability. Raw RNA-seq data and expression tables from ESCC patients have been submitted to the National Center for Biotechnology Information Mutagenesis Assay. Mutagenesis of the PRC2-EZH2 binding site in HERES was (NCBI) Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/) conducted using a Q5 Site-Directed Mutagenesis kit. HERES_NheI_F acccaagctggctagc under accession number GSE130078. LncRNA annotation constructed in this GGCAACCAGCTTGCGTCC and HERES_XbaI_r aaacgggccctctaga TTTAAATGA- study can be downloaded from our website (http://big.hanyang.ac.kr/CASOL/ TAGGGTATTGG were used as the cloning primers to insert the wild-type (WT) catalog.html). HERES cDNA into pcDNA3.1. ACKNOWLEDGMENTS. We thank all Bioinformatics and Genomics (BIG) lab HERES-Mut_AS: members for critical reading and comments. The results shown here are in TATTTACCTGAATATGTCTTGTATTCATTCATGTAAATTA and part based upon data generated by the TCGA Research Network: https:// HERES-Mut_S: TAATTTACATGAATGAATA C AAGACATATTCAGGTAAATA www.cancer.gov/tcga. The TCGA and GTEx data used in this manuscript were primers were used to create HERES-Mut. HERES-Wt and HERES-Mut se- obtained from the database of Genotypes and Phenotypes (dbGaP) through accession numbers phs000178.v10.p8 and phs000424.v6.p1, respectively. This quences were confirmed by Sanger sequencing. work was supported by Korean Health Technology R&D Project, Ministry of Health and Welfare (HI15C3224) and by the Bio and Medical Technology De- velopment Program and the Basic Science Research Program through the Xenograft Assay in Nude Mice. All animal experiments were performed using a National Research Foundation (NRF), funded by the Ministry of Science protocol approved by Institutional Animal Care and Use Committee at the and Information and Communication Technologies (ICT) (grant numbers Catholic University, College of Medicine (2018-0619-01). After transfection 2017M3A9G8084539, 2017R1A2B4006316, and 2018R1A2B2003782).

1. R. Nusse, H. E. Varmus, Wnt genes. Cell 69, 1073–1087 (1992). 25. J. H. Yoon et al., The long noncoding RNA LUCAT1 promotes tumorigenesis by con- 2. P. Polakis, Wnt signaling and cancer. Genes Dev. 14, 1837–1851 (2000). trolling ubiquitination and stability of DNA methyltransferase 1 in esophageal 3. F. Deng, K. Zhou, W. Cui, D. Liu, Y. Ma, Clinicopathological significance of wnt/ squamous cell carcinoma. Cancer Lett. 417,47–57 (2018). β-catenin signaling pathway in esophageal squamous cell carcinoma. Int. J. Clin. Exp. 26. C. Q. Li et al., Integrative analyses of transcriptome sequencing identify novel func- Pathol. 8, 3045–3053 (2015). tional lncRNAs in esophageal squamous cell carcinoma. Oncogenesis 6, e297 (2017). 4. T. Zhan, N. Rindtorff, M. Boutros, Wnt signaling in cancer. Oncogene 36, 1461–1473 27. B. H. You, S. H. Yoon, J. W. Nam, High-confidence coding and noncoding tran- (2017). scriptome maps. Genome Res. 27, 1050–1062 (2017). 5. T. Grigoryan, P. Wend, A. Klaus, W. Birchmeier, Deciphering the function of canonical 28. A. R. Forrest et al.; FANTOM Consortium and the RIKEN PMI and CLST (DGT), A Wnt signals in development and disease: Conditional loss- and gain-of-function mu- promoter-level mammalian expression atlas. Nature 507, 462–470 (2014). tations of beta-catenin in mice. Genes Dev. 22, 2308–2341 (2008). 29. J. W. Nam et al., Global analyses of the effect of different cellular contexts on Cell – 6. H. Clevers, Wnt/beta-catenin signaling in development and disease. 127, 469 480 microRNA targeting. Mol. Cell 53,1031–1043 (2014). (2006). 30. Y. Li et al., Investigation of tumor suppressing function of CACNA2D3 in esophageal 7. M. V. Semenov, R. Habas, B. T. Macdonald, X. He, SnapShot: Noncanonical Wnt sig- squamous cell carcinoma. PLoS One 8, e60027 (2013). Cell naling pathways. 131, 1378 (2007). 31. A. Wanajo et al., Methylation of the calcium channel-related gene, CACNA2D3, is 8. R. T. Moon, A. D. Kohn, G. V. De Ferrari, A. Kaykas, WNT and beta-catenin signalling: frequent and a poor prognostic factor in gastric cancer. Gastroenterology 135, 580– Nat. Rev. Genet. – Diseases and therapies. 5, 691 701 (2004). 590 (2008). 9. J. N. Anastas, R. T. Moon, WNT signalling pathways as therapeutic targets in cancer. 32. A. M. Wong et al., Characterization of CACNA2D3 as a putative tumor suppressor Nat. Rev. Cancer – 13,11 26 (2013). gene in the development and progression of nasopharyngeal carcinoma. Int. J. 10. V. Conteduca et al., Barrett’s esophagus and esophageal cancer: An overview. Int. J. Cancer 133, 2284–2295 (2013). Oncol. 41, 414–424 (2012). 33. H. Zou et al., Aberrant methylation of secreted frizzled-related protein genes in 11. K. J. Napier, M. Scheerer, S. Misra, Esophageal cancer: A review of epidemiology, esophageal adenocarcinoma and Barrett’s esophagus. Int. J. Cancer 116, 584–591 pathogenesis, staging workup and treatment modalities. World J. Gastrointest. On- (2005). col. 6, 112–120 (2014). 34. M. T. Chung et al., SFRP1 and SFRP2 suppress the transformation and invasion abilities 12. S. Ohashi et al., Recent advances from basic and clinical studies of esophageal squa- of cervical cancer cells through Wnt signal pathway. Gynecol. Oncol. 112, 646–653 mous cell carcinoma. Gastroenterology 149, 1700–1715 (2015). (2009). 13. K. Higuchi et al., Current management of esophageal squamous-cell carcinoma in 35. E. Viré et al., The Polycomb group protein EZH2 directly controls DNA methylation. Japan and other countries. Gastrointest. Cancer Res. 3, 153–161 (2009). Nature 439, 871–874 (2006). 14. Y. Miyawaki et al., Efficacy of docetaxel, cisplatin, and 5-fluorouracil chemotherapy 36. H. Cedar, Y. Bergman, Linking DNA methylation and histone modification: Patterns for superficial esophageal squamous cell carcinoma. Dis. Esophagus 30,1–8 (2017). Nat. Rev. Genet. – 15. M. W. Wiedmann, J. Mössner, New and emerging combination therapies for esoph- and paradigms. 10, 295 304 (2009). et al ageal cancer. Cancer Manag. Res. 5, 133–146 (2013). 37. T. Kojima ., Decreased expression of CXXC4 promotes a malignant phenotype in Oncogene – 16. X. Kang et al., Personalized targeted therapy for esophageal squamous cell carci- renal cell carcinoma by activating Wnt signaling. 28, 297 305 (2009). noma. World J. Gastroenterol. 21, 7648–7658 (2015). 38. M. Guttman, J. L. Rinn, Modular regulatory principles of large non-coding RNAs. Nature – 17. C. P. Ponting, P. L. Oliver, W. Reik, Evolution and functions of long noncoding RNAs. 482, 339 346 (2012). et al Cell 136, 629–641 (2009). 39. X. Wang ., Targeting of polycomb repressive complex 2 to RNA by short repeats Mol. Cell – 18. I. Ulitsky, D. P. Bartel, lincRNAs: Genomics, evolution, and mechanisms. Cell 154,26–46 of consecutive guanines. 65, 1056 1067.e5 (2017). (2013). 40. Y. Lubelsky, I. Ulitsky, Sequences enriched in Alu repeats drive nuclear localization of Nature – 19. M. Huarte, The emerging role of lncRNAs in cancer. Nat. Med. 21, 1253–1261 (2015). long RNAs in human cells. 555, 107 111 (2018). et al 20. J. F. Collins, Long noncoding RNAs and hepatocellular carcinoma. Gastroenterology 41. Y. Long ., Conserved RNA-binding specificity of polycomb repressive complex 2 is 148, 291–294 (2015). achieved by dispersed amino acid patches in EZH2. eLife 6, e31558 (2017). 21. A. M. Schmitt, H. Y. Chang, Long noncoding RNAs in cancer pathways. Cancer Cell 29, 42. V. Zarkou, A. Galaras, A. Giakountis, P. Hatzis, Crosstalk mechanisms between the 452–463 (2016). WNT signaling pathway and long non-coding RNAs. Noncoding RNA Res. 3,42–53 22. J. H. Yuan et al., A long noncoding RNA activated by TGF-β promotes the invasion- (2018). metastasis cascade in hepatocellular carcinoma. Cancer Cell 25, 666–681 (2014). 43. Cancer Genome Atlas Research Network et al., Integrated genomic characterization 23. M. M. Ali et al., PAN-cancer analysis of S-phase enriched lncRNAs identifies oncogenic of oesophageal carcinoma. Nature 541, 169–175 (2017). drivers and biomarkers. Nat. Commun. 9, 883 (2018). 44. GTEx Consortium, The Genotype-tissue expression (GTEx) project. Nat. Genet. 45, 24. Y. Wu et al., Up-regulation of lncRNA CASC9 promotes esophageal squamous cell 580–585 (2013). carcinoma growth by negatively regulating PDCD4 expression through EZH2. Mol. 45. National Research Council, Guide for the Care and Use of Laboratory Animals (Na- Cancer 16, 150 (2017). tional Academies Press, Washington, DC, ed. 8, 2011).

10 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1912126116 You et al.